Respiratory Monitoring During Physical Activities With a Multi-Sensor Smart Garment and Related Algorithms

Unobtrusive and wearable devices are gaining large acceptance in the continuous monitoring of physiological parameters. Among the five vital signs, respiratory rate (<inline-formula> <tex-math notation="LaTeX">${f}_{R}$ </tex-math></inline-formula>) can be used to detect physiological abnormalities and health status changes. The purpose of this work was to investigate the performances of a multi-sensor smart garment in estimating the <inline-formula> <tex-math notation="LaTeX">${f}_{R}$ </tex-math></inline-formula> during walking and running activities. Bespoke algorithms have been implemented to retrieve <inline-formula> <tex-math notation="LaTeX">${f}_{R}$ </tex-math></inline-formula> values from raw data. Experiments were carried out on ten male volunteers during walking and running activities at selected speeds controlled by a treadmill (i.e., from 1.6 km<inline-formula> <tex-math notation="LaTeX">$\cdot \text{h}^{-{1}}$ </tex-math></inline-formula> to 8.0 km<inline-formula> <tex-math notation="LaTeX">$\cdot \text{h}^{-{1}}$ </tex-math></inline-formula>). Data were analysed in both frequency and time domains. In the frequency domain, <inline-formula> <tex-math notation="LaTeX">${f}_{R}$ </tex-math></inline-formula> was analyzed considering a time window of 20 s. The 97% of <inline-formula> <tex-math notation="LaTeX">${f}_{R}$ </tex-math></inline-formula> estimated by the garment agreed with the reference (i.e., flowmeter) values in the range ±3 breaths per minute (bpm). In the time domain, breath-by-breath <inline-formula> <tex-math notation="LaTeX">${f}_{R}$ </tex-math></inline-formula> analysis was carried out. The garment performance was evaluated in terms of mean absolute error (MAE), standard error (SE), mean percentage error (mean <inline-formula> <tex-math notation="LaTeX">$\%{E}[{i}]$ </tex-math></inline-formula>) and by the Bland-Altman analysis. Good agreement with the reference device was testified by low MAE (<1.86 bpm), SE (<0.21 bpm), mean <inline-formula> <tex-math notation="LaTeX">$\%{E}[{i}]$ </tex-math></inline-formula> (<2.83 %), and by the Bland-Altman analysis (Mean of Differences = 0.22 bpm, Limits of Agreement = 6.06 bpm). Summing up, the garment based on six sensing elements and related bespoke algorithms are able to provide robust information about <inline-formula> <tex-math notation="LaTeX">${f}_{R}$ </tex-math></inline-formula> on both average and breath-by-breath bases even during physical activities.